Discussion: Scalability group's stance on future Redis deployments (k8s or VMs)
This issue's purpose is to house discussion points and conclusions on Scalability's stance for future Redis deployments. Originating from: #2334 (comment 1379570972).
Questions to answer:
- Should we be K8s-first or VMs-first?
- What actionable to support that stance? (Quoting Bob:
If it's kubernetes, we should spin up new instances there, if they fit, but we need to be conscious of unknowns that are still there. If it's VMs, we should do it there and perhaps with each iteration try to pay back some of the debt in the chef repo, making it less of a pain.
)
Some existing considerations (feel free to edit and add to this list):
- toil in spinning-up new instances
- performance overheads
- ease of debugging/observability
- maintenance toil
Past discussions & efforts:
Edited by Sylvester Chin